Physiological blushing in social anxiety disorder patients with and without blushing complaints: Two subtypes?

Biological Psychology 81 (2009) 86–94

Contents lists available at ScienceDirect

Biological Psychology journal homepage: www.elsevier.com/locate/biopsycho

Physiological blushing in social anxiety disorder patients with and without blushing complaints: Two subtypes? Marisol J. Voncken a,*, Susan M. Bo¨gels b a b

Maastricht University, Clincical Psychological Science, P.O. Box 616, 6200 MD Maastricht, The Netherlands University of Amsterdam, The Netherlands

A R T I C L E I N F O

A B S T R A C T

Article history: Received 1 August 2008 Accepted 13 February 2009 Available online 27 February 2009

This study investigates whether social anxiety disorder (SAD) patients with blushing complaints show heightened physiological blushing and arousability in social situations than SAD patients without blushing complaints and healthy controls. SAD blushers (n = 32), SAD non-blushers (n = 34), and healthy controls (n = 25) conducted two social tasks. The physiological responses cheek and forehead blood flow, cheek temperature, and skin conductance were recorded, as well as confederates-observed blushing. The SAD blushers showed more physiological blushing (cheek temperature and blood flow) than SAD nonblushers and observers detected this difference. This finding was also present in comparison to the controls, except for blood flow. For blood flow SAD blushers and controls did not differ but SAD nonblushers showed a ‘suppressed response’: a smaller cheek blood flow increase during the interaction and no recovery compared to the other groups. Furthermore, on skin conductance no differences between groups were observed. Discussed is to what extent SAD blushers and SAD non-blushers represent two qualitative distinct subgroups of SAD. ß 2009 Elsevier B.V. All rights reserved.

Keywords: Social phobia Social anxiety disorder Blushing Skin temperature Blood flow Skin conductance

Patients with social anxiety disorder (SAD) fear negative evaluation by others. In 1872 Darwin stated that it is ‘‘the thinking about what other people can think of us which excites a blush’’ (Darwin, 1955, p. 325). As a result, people who think that others evaluate them negatively would be more prone to blush. Indeed, blushing is a frequently reported anxiety symptom of patients with SAD (Edelmann, 1990). A proportion of SAD patients even develop a blushing phobia or erythrophobia (Bo¨gels and Stein, 2009; Bo¨gels et al., 1997; Edelmann, 1990; Mersch et al., 1992; Scholing and Emmelkamp, 1993). The phobic properties of blushing may result from several features. First, as blushing is an overt manifestation of responses regulated by the autonomic nervous system, it is not under subjects’ voluntary control. Second, in contrast to other physiological anxiety symptoms like heart palpitations, blushing can be noticed by others. Third, blushing serves to increase self-focused attention (Bo¨gels and Mansell, 2004), thereby making the blusher more aware of own anxiety symptoms, and, as a result, increase the fear of negative evaluation by others because of showing anxiety symptoms. Finally, as concomitants of blushing are to avoid gaze and bend the head (Leary and Meadows, 1991), blushing may interfere with social behaviour.

* Corresponding author. E-mail addresses: [email protected] (M.J. Voncken), [email protected] (S.M. Bo¨gels). 0301-0511/$ – see front matter ß 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.biopsycho.2009.02.004

Although a characteristic of normal human functioning is the capacity to blush, individuals differ in their blushing propensity and intensity (Leary et al., 1992). Therefore, one of the etiological markers of the development of SAD might well be vulnerability for more frequent and/or intense blushing. If a heightened blushing propensity is indeed a marker for SAD, it could be hypothesized that individuals with SAD blush more. However, studies so far have produced conflicting results. Blushing is measured in various ways: temperature of the cheek and ear lobes, or photoplethysmograph recordings of cheek or forehead. Researchers measure mean blushing level, and/or blushing responsiveness in socially challenging situations. Four studies found evidence for a heightened physiological blush response in SAD. This evidence consists of a higher mean blushing level in socially anxious individuals (Bo¨gels et al., 2002), and a greater blushing response in both socially anxious students (Hofmann et al., 2006), people seeking treatment for blushing complaints (Drummond, 2001) or with SAD patients recruited by advertisements (Gerlach et al., 2001). It is worth noting that Gerlach et al. (2001) only recorded differences in physiological blushing while participants watched an embarrassing video, and not during a speech and an interaction task. Four other studies did not find evidence for social anxiety to be marked by heightened blush responses. These studies included analogue populations with high versus low blushing fearful or high versus low blushing propensity students (Drummond et al., 2007; Mulkens et al., 1997, 1999) and a population of patients with SAD recruited through

M.J. Voncken, S.M. Bo¨gels / Biological Psychology 81 (2009) 86–94

advertisements versus normal controls (Edelmann and Baker, 2002). In other words, there is conflicting evidence as to whether individuals with SAD blush more frequently or more intensely than normal controls. One explanation for the lack of clarity concerning whether or not SAD is characterized by blushing, is that not all people with SAD blush more than those without SAD. There might be large variability within the group of individuals with SAD with respect to blushing propensity and blushing intensity. Only one study (Gerlach et al., 2001), examined whether SAD patients (n = 15) with blushing as their primary complaint differed in their psychophysiological responses from other patients with SAD (n = 15). They were not able to identify a different blushing response, as measured with a cheek photoplethysmograph, to their social tasks (e.g., watching an embarrassing video, conversation and speech) between these two groups of SAD patients. However, this might be due to power problems. Moreover, it is worth noting that their patients were recruited by advertisements and were not seeking help through regular mental health care. Gerlach et al. (2001) did, however, find SAD patients with blushing complaints to show higher heart rate during their three social tasks than SAD patients without these complaints. The latter finding is in line with the findings of Laederach Hofman et al. (2002), who found that erythrophobia patients have higher heart rates during a mental stress task than healthy subjects. However, they did not assess whether their patients fulfilled the diagnosis of SAD and did not compare their sample to SAD patients without blushing complaints. Hofmann et al. (2004) argue that rather than blushing per se, one subtype of SAD would be a fearful subtype that reacts with a more intense general psychophysiological response to social stress. Therefore, SAD patients with blushing complaints might suffer from heightened general arousability rather than heightened vascular blushing. In conclusion, although SAD is characterized by blushing, so far only a few studies have been able to detect differences between socially anxious and non-socially anxious individuals in physiological blushing measurements such as cheek temperature and cheek and forehead blood flow. This lack of consistency in the literature might reflect the heterogeneity of SAD patients and it seems necessary to differentiate SAD patient with and without blushing complaints. However, the only one study that did differentiate between these two groups of SAD patients did not find differences in physiological blushing responses. Still, it might be that only those SAD patients who complain about blushing, can be characterized by more (intense) physiological blushing or by a general heightened psychophysiological response to social stress. This study aims to investigate physiological blushing and arousability in patients with SAD with and without blushing complaints that were seeking help in regular mental health care,

87

and in healthy control participants. To that end, SAD patients characterized as blushers (n = 32) and non-blushers (n = 34), and normal controls (n = 25) conducted both a speech and a conversation. Two social tasks were included to increase reliability of the assessments. The physiological responses cheek and forehead vasodilatation, cheek temperature, and skin conductance were recorded during baseline, interaction, and recovery. It was hypothesized that SAD patients with blushing complaints have a stronger blush response or blush more intensely and have a general heightened arousability than SAD patients without blushing complaints and healthy subjects in social situations. 1. Method 1.1. Participants The SAD patients were referred to the community mental health centre of Maastricht, The Netherlands, for out-patient treatment. Diagnoses were determined with the Structured Clinical Interview for DSM-IV Axis-I disorders (SCID-I, First et al., 1997) by trained therapists. Of the 81 SAD patients that were approached, 66 (81%) participated. Reasons for non-participation were: in five patients their confidentiality was at risk, as they were students or employees at the university and could be recognized by the student-confederates, three refused because of the time investment and six because of anxiety. Two patients, overwhelmed by anxiety, dropped out during the assessment. Except for two patients, all patients fulfilled the criteria of generalized SAD (97%). The SAD patients were divided into two groups based on an extra question added to the SCID-I interview: whether they ‘suffered from blushing during social interactions’. Those who responded affirmatively (n = 32) are regarded as the blush group whereas those who responded not affirmatively (n = 34) are regarded as the non-blush group. The normal control group (n = 25) was recruited from a general list of people willing to participate in research. None of them fulfilled criteria of either SAD or depression, as assessed with the SCID-I. All participants completed the Social Phobia and Anxiety Inventory (SPAI; Turner et al., 1989; Dutch validation by Bo¨gels and Reith, 1999). Characteristics of the participants are depicted in Table 1. It was checked whether the two SAD groups differed on variables that could influence physiological responses. First, no differences in symptom severity were found between the two SAD groups on the SPAI, t(56) = 0.8, p = .41, total psychopathology score on the Symptoms Check List (SCL-90), t(56) = 0.4, p = .68, and number of criteria of avoidant personality rated by trained therapists using the SCID-II, t(56) = 0.9, p = .38. However, the groups did differ in number of life-time mood disorders (SAD blushers 37.5% versus SAS non-blushers 61.8%, x2(1) = 3.9, p = .049. Second, although no evidence is present (see Leary et al., 1992) it is generally believed that age and sex might affect blush symptoms. No differences were found between the two SAD groups on age, t(56) = 1.1, p = .29. However, there was a sex difference between the groups, x2(1) = 5.5, p = .02. The SAD blushers group consisted of more women than men (19 women and 12 men), whereas, the SAD non-blushers group consisted of more men than women (9 women and 20 men). The control group did not differ in gender distribution from the SAD blushers x2(1) = 0.3, p = .62, but the SAD non-blushers consisted of a higher percentage of men than the control group, x2(1) = 4.2, p = .04. Due to equipment failure for eight participants (one SAD blusher, five SAD nonblushers and two controls) all physiological data were lost. In addition, during recording for six participants the equipment partly failed and some of their data were lost (five SAD non-blushers: one cheek temperature, two skin conductance, two both cheek temperature as well as cheek and forehead blood flow, and one normal control: skin conductance).

Table 1 Characteristics of the participants.

Age Sex Education levela SPAI, social phobia subscale Avoidant personalityb SCL-90 total score Life-time depressionc Blush item of SPAI (item 32) Number of blushing—fearful social situations Self-rating blushing Observer rating blushing

SAD blushers n = 32 (S.D.)

SAD non-blushers n = 34 (S.D.)

Normal controls n = 25 (S.D.)

30.3 (8.7) 38% male a 8.0 (2.1) 129.9 (31.8) a 3.6 (2.1) a 183 (61) a 37.5% a 5.1 (1.2) a 2.0 (2.3) a 5.4 (2.1) a 4.3 (1.5) a

33.1 (10.1) 71% male b 8.0 (2.3) 124.0 (28.2) a 3.1 (2.1) a 187 (61) a 61.8% b 3.3 (1.5) b 0 (0) b 3.3 (1.7) b 2.8 (1.2) b

32.5 (11.3) 52% male a 7.7 (1.7) 57.2 (28.7) b – – – 2.8 (1.5) b – 3.3 (1.8) b 2.8 (1.3) b

Note: Means with different letters differ significantly (p < .05, two-tailed). a On a 11-point scale, from 1 = no completed education, 2 = elementary school, to 11 = masters degree. b Number of criteria of avoidant personality disorder assessed by a SCID II interview. c Life-time mood disorder assessed by a SCID I interview.

88

M.J. Voncken, S.M. Bo¨gels / Biological Psychology 81 (2009) 86–94

1.2. Procedure Before starting participants were inquired about medication, caffeine and alcohol use as these factors could influence the physiological assessment. None of them took anxiolytic medication less than 4 days previous to the assessment or consumed caffeine or alcohol less than 1 h previous to the assessment (as instructed). After being attached to the equipment for the physiological recordings, participants got adjusted to the equipment and the laboratory temperature for 20 min. Prior to each task baseline level of physiological arousal was assessed during a 2.5-min baseline. The instruction for the speech was as follows: ‘‘We would like you to give a presentation of 5 min in front of an audience of two persons. The audience will listen and not react and will judge your social skills afterwards. Please give a presentation about (1) the city/town where you grew up, or (2) a holiday destination you have been or want to go to.’’ The speech topics were randomized. Participants received a 2.5-min speech preparation period. The instruction for the conversation was: ‘‘We would like you to have a conversation with two persons, with the purpose of getting to know each other. It is up to you to start the conversation and keep it going.’’ Unlike for the speech, participants did not receive preparation time. Order of the tasks, speech and conversation, as well as topic of the speech were randomly assigned. Participants were explained that the confederates did not know whether they were patients or not and to keep the confederates blind they were asked not to talk about treatment or the study itself. After each task participants had a recovery period of 2.5-min and were instructed to sit quietly and relax. The temperature of the laboratory was kept a few degrees lower than the outside temperature. After the recovery phase the participants filled in a questionnaire concerning their social performance. 1.3. Confederates To create a homogeneous group, all confederates were undergraduate students. For each assessment four confederates were needed: one male and one female for each of the two social tasks. During the course of this study a total of 58 confederates participated. They were trained to behave according to instructions ¨ st et al. (1981). That is, they were to remain based on Boone et al. (1999) and O neutral during the speech and the conversation. During the conversation they were instructed to leave the burden of the conversation with the participant, were not allowed to change the subject of the conversation and were only allowed to take an initiative when the participant was silent for 7 s. Furthermore, they were constrained in their responses to three pieces of information per answer. After the social task the confederates rated the social performance of the participants. 1.4. Physiological measures Multi-channel psychological recordings were made with a Vitaport III System. The vasodilatation of the blood vessels (referred to as blood flow) in cheek and forehead skin was recorded from a HP model 15230A plethysmograph transducer that was modified in such a way that it was dc coupled rather than ac coupled, and sampled at a frequency of 1024 Hz (cf. Shearn et al., 1990; Mulkens et al., 1999). To assess cheek blood flow the infrared probe for cheek was placed about 5 cm below the right corner of the right eye. For forehead blood flow the probe was placed in the middle of the forehead 2 cm above the eyebrow. Cheek temperature was assessed by means of the Vitaport temperature transducer that was dc coupled. Skin conductance level was recorded from two Ag–AgCl electrodes (8 mm diameter), placed on the medial phalanges of the middle finger and ring finger of participants’ nondominant hand, using the method of constant voltage (0.5 V). The electrodes were filled with an isotonic paste and connected to the Vitaport system. Both temperature and skin conductance were sampled with a frequency of 16 Hz. All physiological parameters were analyzed off-line by means of a specifically designed computer program. For each phase in the experiment a mean response for each of the four physiological parameters was calculated. 1.5. Self- and observer-rating for blushing After each social task, both participants as well as observers rated the social performance of the participants with a 24- or 27-item questionnaire on a 9-point Likert scale. These data are presented in another study (Voncken and Bo¨gels, 2008). For the current study two items specifying frequency and intensity of blushing were used (e.g., ‘How often did the participant blush?’ ‘How intense did the participant blush?’). These items showed good reliability (alpha’s: self-rating: .91; observer rating: .87). 1.6. Data preparation First, all variables were checked on normality. The skin conductance data were not normally distributed. For the analyses that require normally distributed data (e.g., ANOVAs and t-tests) the skin conductance data were transformed by a square root transformation. Second, we inspected whether the correlations between the speech and conversation were high for each of the physiological measures. We concluded that these correlations were high (for baseline between .61 and .94; for

task between .84 and .92; for recovery between .79 and .95) and therefore we combined the assessments of the two social tasks for each of the physiological assessments. Thus, we composed for each of the four physiological measures three variables: baseline, social task, and recovery. Third, Drummond (2001) asserts that both cheek and forehead blood flow reflect the same blushing responses. We investigated whether this was also true for our data by inspecting if the correlations between the two measures were high and if the response pattern of both indices were highly similar across the three groups (see Section 2). 1.7. Data analyses To analyze the main research question about differences in physiological responses to the social tasks between the three groups, repeated measures ANOVA was used with phase as within subject variable (baseline, social task, recovery) and as between subject variable the three groups (SAD blushers, SAD non-blushers and controls). The authors chose to analyze linear and quadratic trends, an analysis possible with three data points. For the current physiological data both types of interactions are of interest. The quadratic interaction is of importance because participants generally show an increase from baseline to task and a decrease from task to recovery. In other words, the data normally show a peak at the moment of the task. Differences between our three groups were expected for the shape of this peak. That is, we expected to SAD blushers to show a more peaked pattern and the SAD non-blushers and normal controls a more flattened pattern, indicating a stronger blushing response in the SAD blushers than in the other two groups. On the other hand we expected the SAD blushers to show a steeper linear pattern than the other two groups, indicating that this group would not recover as well from the physiological increase during the task than the other two groups (see Drummond et al., 2007 for such findings). We first analyzed how our four different physiological measures correlated with each other to understand whether and how they are related. As physiological measures can be confounded by a variation of variables such as thickness of the skin, light angle or placement of the probe on the skin, it is generally preferred (see Drummond, 2001) to use variation in response from baseline to task instead of mean levels of physiological assessment. Thus, two variables were constructed, (1) a ‘response rating’ in which the baseline assessment was subtracted from the assessment during the social task and (2) a ‘recovery rating’ in which the recovery assessment was subtracted from the assessment during the social task. Correlations between the four physiological indices for these variables were calculated. Second, we were interested how self-report and observations of blushing related to the physiological measures. Prior studies (see for instance Drummond et al., 2007) concluded that in their student samples people are not able to detect their blushing response and it is worthwhile to investigate whether this also holds for patients. As the collected self-report and observer ratings referred to blushing experiences during the social task, correlations between the physiological ‘response ratings’ (social task minus baseline) and self-report and observers ratings were calculated. Third, as the SAD groups differed in gender distribution, we investigated the effect of sex on the physiological measures. Therefore, we conducted repeated measures ANOVAs with as between subject variable gender and as within subject variable phase (baseline, social task and recovery). For cheek and forehead blood flow no main effect of sex, resp. F(1,78) = 0.1, p = .7 and F(1,78) = 0.8, p = .4, and no interaction effect between sex and phase, resp. F(2,79) = 1.0, p = .4 and F(2,79) = 0.3, p = .7, were found. Thus, sex did not affect the physiological responses on blood flow. However, for cheek temperature and skin conductance main effects, resp. F(1,79) = 14.0, p = .000, j2 = .15 and F(1,77) = 7.9, p = .006, j2 = .09, occurred. Inspection of the means revealed that men had a higher cheek temperature and skin conductance level than women. Next to the main effect also interaction effects emerged. That is, a linear interaction for cheek temperature, F(1,79) = 13.6, p = .000, j2 = .15, occurred, indicating a stronger increase of temperature in women than in men. For skin conductance a quadratic effect, F(1,77) = 9.1, p = .004, j2 = .11,

Table 2 Correlations between the four physiological indices for responsiveness and recovery. Physiological indices

Responsiveness to the social Forehead blood flow Cheek temperature Skin conductance

Cheek blood flow taska .25* .30* .08

Forehead blood flow

Cheek temperature

.31* .12

.01

.11 .04

.12

b

Recovery from the social task Forehead blood flow .24* Cheek temperature .05 Skin conductance .09 a b *

Calculated by subtracting baseline assessments from social task assessments. Calculated by subtracting recovery assessments from social task assessments. Correlation is significant at the 0.05 level (two-tailed).

M.J. Voncken, S.M. Bo¨gels / Biological Psychology 81 (2009) 86–94

89

Table 3 Correlations between the four physiological measuresa and self- and observer-rated blush for each of the three groups. Physiological indices

Normal controls Self-report blush

Cheek blood flow Forehead blood flow Cheek temperature Skin conductance a * +

*

.53 .01 .38+ .02

SAD blushers Observers-rated blush .26 .05 .11 .10

Self-report blush .34 .01 .19 .02

+

SAD non-blushers Observers-rated blush .16 .02 .26 .12

Self-report blush .12 .11 .03 .04

Observers-rated blush .01 .06 .15 .05

Calculated by subtracting baseline assessments from social task assessments. Correlation is significant at the 0.05 level (two-tailed). Correlation is borderline significant (p < .10, two-tailed).

occurred, indicating that women were more responsive than men from baseline to social task, t(79) = 2.8, p = .007, d = .63. See footnotes 3 and 4 for the ANOVAs concerning cheek temperature and skin conductance in which gender was included as covariate. It is worth noting that Drummond and Lim (2000) did not find forehead blood flow or cheek temperature to be influenced by gender. All post hoc t-tests used to unravel interactions were Bonferroni corrected by multiplying the p-value by the number of tests used. The data for the physiological measures that were lost due to equipment failure are represented in the degrees of freedom of the subsequent analyses.

found cheek blood flow to correlate with self- and observerrated blush in our normal control sample. No such relations were found in both the SAD groups. Self- and observer-rated blush did not correlate with forehead blood flow, cheek temperature and skin conductance level in any of the three groups.

2. Results

Drummond (2001) argues that both cheek and forehead blood flow reflect the same blushing responses and our data indeed showed cheek and forehead blood flow to correlate significantly (see above and Table 3). Moreover, the effects from the ANOVA analyses of both blood flow measures were highly identical.1 Therefore, to increase the reliability and, therefore, the power of our analyses we decided to construct one blood flow variable by averaging the forehead and cheek blood flow assessments for each of the three assessments times (baseline, social task and recovery). First, the repeated measures ANOVAs were run with the SAD groups combined versus normal controls as between group variable and with phase (baseline, social task, recovery) as within group variable. Second, the same analyses were run with the three different groups (SAD blushers, SAD non-blushers, normal controls) as between group variable. See Table 4 for the mean levels for the four physiological assessments in each phase of the experiment for the three different groups. See Fig. 1 for graphs of the three physiological indices, cheek temperature, combined blood flow of cheek and forehead, and skin conductance in each phase of the experiment for the different groups.

2.1. Validity of the classification of SAD with and without blushing complaints To check whether the classification of the SAD blushers versus SAD non-blushers was reliable, various blush indices were used (see Table 1 for mean and standard deviations). In line with expectations, the SAD blushers scored higher than the SAD nonblushers on the blush item of the SPAI (item 32 ‘I experience the following during a social situation: blushing’), t(48) = 4.7, p = .000, d = 1.3). For treatment purposes patients formulated five idiosyncratic fearful social situations. These situations included more blush situations for the SAD blushers than for the SAD nonblushers, t(63) = 5.1, p = .000, d = 1.2). During the social tasks the SAD blushers rated themselves as blushing more than the SAD nonblushers, t(64) = 4.4, p = .000, d = 1.1. Not only did the SAD blushers rate themselves as blushing more than the SAD non-blushers, also the confederates identified that the SAD blushers blushed more than the SAD non-blushers, t(64) = 4.6, p = .000, d = 1.1. These results support the validity of the classification of the blushers versus the non-blushers within the SAD group. It is worth noting that the observers also detected the SAD blushers to blush more than the normal controls, t(54) = 3.8, p = .000, d = 1.0, but did not see differences between the SAD non-blushers and the normal controls. 2.2. Correlations between the physiological measures See Table 2 for the correlations. For responsiveness (social task minus baseline) the three blushing indices (cheek and forehead blood flow and cheek temperature) correlated significantly. In contrast, skin conductance level did not correlate with the blushing measures. For recovery (social task minus recovery) only cheek and forehead blood flow correlated and no correlation between temperature and blood flow emerged. Again, skin conductance level did not correlate with the blushing measures. 2.3. Correlations of the physiological measures with self- and observer-ratings See Table 3 for correlations. In contrast to previous studies (see the overview of Drummond et al., 2007) we

2.4. Main analyses

2.4.1. Phase effect Comparing the combined SAD group versus the normal controls, a quadratic phase effect was found for each of the three physiological measures (blood flow: F(1,80) = 45.1, p = .000, j2 = .36; cheek temperature: F(1,81) = 6.1, p = .016, j2 = .07; skin conductance level: F(1,79) = 98.3, p = .000, j2 = .55). The exact same pattern was found in the analyses in which the three different groups were included in the analyses.2 Post hoc paired t-tests revealed that, as expected, participants’ physio1 The results from the ANOVA analyses with the three groups for each of the blood flow indices—phase effect: cheek blood flow: F(1,79) = 47.1, p = .000; forehead blood flow: F(1,79) = 83.2, p = .000, group effect: cheek blood flow: F(1,79) = 1.4, p = .25; forehead blood flow: F(1,79) = 0.2, p = .78, linear interaction effect: cheek blood flow: F(1,79) = 0.3, p = .75; forehead blood flow: F(1,79) = 1.3, p = .27, and quadratic interaction effect: cheek blood flow: F(2,78) = 3.9, p = .02.; forehead blood flow: F(2,79) = 4.1, p = .020. 2 For the ANOVAs with the SAD blushers, SAD non-blushers and normal controls as between subject variables the following quadratic phase effects were found— blood flow: F(1,78) = 51.1, p = .000, j2 = .39; cheek temperature: F(1,80) = 7.3, p = .008, j2 = .08; skin conductance level: F(1,78) = 99.9, p = .000, j2 = .56. In addition, the following linear phase effects appeared: blood flow: F(1,79) = 25.4, p = .000, j2 = .24; cheek temperature: F(1,80) = 74.2, p = .000, j2 = .48.

M.J. Voncken, S.M. Bo¨gels / Biological Psychology 81 (2009) 86–94 2.42 (0.68) 2.60 (0.67) 2.45 (0.65) 2.53 (0.77) 2.60 (0.74) 2.52 (0.75) 2.58 (0.54) 2.65 (0.50) 2.51 (0.51) 33.6 (1.0) 33.8 (1.1) 34.0 (1.1) 33.5 (1.1) 33.6 (1.1) 33.8 (1.1)

Fig. 1. Depicted are the three physiological indices, cheek temperature, blood flow, and skin conductance, during baseline, social task and recovery for the SAD blushers, SAD non-blushers and normal control group.

4.68 (3.74) 6.63 (3.61) 5.15 (4.28) Square root transformation. a

5.19 (4.14) 5.84 (4.57) 6.06 (4.64) Baseline Social task Recovery

4.47 (3.98) 6.39 (5.04) 5.51 (5.15)

2.84 (2.01) 4.67 (4.07) 4.12 (3.71)

4.00 (4.44) 5.70 (5.31) 4.27 (5.57)

4.17 (3.65) 5.20 (4.34) 5.34 (4.80)

SAD non-blushers n = 29 (S.D.) Normal controls n = 23 (S.D.) SAD nonblushers n = 28 (S.D.) SAD blushers n = 31 (S.D.)

Normal controls n = 23 (S.D.)

SAD blushers n = 31 (S.D.)

SAD nonblushers n = 28 (S.D.)

34.1 (1.4) 34.4 (1.3) 34.6 (1.2)

SAD non-blushers n = 28 (S.D.) SAD blushers n = 31 (S.D.) SAD blushers n = 31 (S.D.)

Normal controls n = 23 (S.D.)

Skin conductance levela Cheek temperature Forehead blood flow Cheek blood flow

Table 4 Means and standard deviations of the four physiological indices of baseline, social task and recovery for the SAD blushers, SAS non-blushers and normal control group.

Normal controls n = 22 (S.D.)

90

logical level increased for all measures from baseline to task (blood flow: t(81) = 7.7, p = .000, d = 1.7; cheek temperature: t(82) = 7.7, p = .000, d = 1.7; skin conductance: t(80) = 6.3, p = .000, d = 1.4) and decreased from task to recovery (blood flow: t(81) = 4.0, p = .000, d = 0.9; cheek temperature: t(82) = 7.3, p = .000, d = 1.6; skin conductance: t(80) = 10.6, p = .000, d = 2.4). In addition, the physiological indices, except for skin conductance, showed a linear phase effect (see the footnote 2) (blood flow: F(1,80) = 21.0, p = .000, j2 = .21; cheek temperature: F(1,81) = 62.8, p = .000, j2 = .43). That is, the physiological levels increased over time, with higher physiological levels during recovery than during baseline (blood flow: t(81) = 5.1, p = .000, d = 1.3; cheek temperature: t(82) = 8.6, p = .000, d = 1.9). 2.4.2. Group effect No main effects for group were found for the three physiological indices comparing the combined SAD groups with the normal controls. However, when the three different groups were included

M.J. Voncken, S.M. Bo¨gels / Biological Psychology 81 (2009) 86–94

in the analyses cheek temperature3 did show a main effect for group, F(2,80) = 3.2, p = .047, j2 = .07. That is, SAD blushers had an overall higher cheek temperature than the SAD non-blushers, F(1,58) = 5.5, p = .022, j2 = .09. The overall cheek temperature of the normal control group was in between the two SAD groups. The difference between the normal control group and the SAD blushers was borderline significant, F(1,52) = 2.8, p = .099, j2 = .05, and no difference was found with the SAD non-blushers, F(1,50) = 0.4, ns. 2.4.3. Interaction effect With regard to cheek temperature no interaction effects between phase and group were found; neither for the analyses in which the combined SAD group was compared to the normal controls as for the analyses in which the three groups were included as between subject variable. As for blood flow, no interaction effects appeared in the analyses with the combined SAD group versus the normal controls, however, with the three groups as between subject variables, no linear but a significant quadratic interaction effect emerged, F(2,79) = 6.4, p = .003, j2 = .14. To unravel the quadratic interaction, the repeated measures ANOVA was ran again with only the two SAD groups as between variable (SAD blushers versus SAD non-blushers). This analysis showed a quadratic interaction effect, F(1,57) = 12.8, p = .001, j2 = .18, which indicated that the SAD blushers compared to the SAD non-blushers showed a greater increase from baseline to social task, t(57) = 2.4, p = .02, d = .64, and a greater decrease from social task to recovery, t(57) = 3.8, p = .005, d = 1.0. Then, each of the two SAD groups was compared separately to the normal control group. Against expectations, no interaction effects were found between the SAD blushers and the normal control group. Surprisingly, between the SAD non-blushers and the controls the quadratic interaction was significant F(1,49) = 7.6, p = .008, j2 = .14. This interaction indicated that the normal controls were more responsive from baseline to social task, t(49) = 2.2, p = .03, d = .63, as well as from social task to recovery, t(49) = 2.8, p = .008, d = .80, than the SAD non-blushers. In other words, the SAD nonblushers showed a flatter blush response pattern than both the SAD blushers and the normal control group. With regard to skin conductance level4, the analysis with the combined SAD group versus normal controls showed both a 3 As cheek temperature showed an effect for gender the analysis were run again with sex included as covariate. Overall the results remained the same. That is, again the main effect for group appeared, F(2,79) = 6.2, p = .003, j2 = 14. The SAD blushers had an overall higher cheek temperature than the SAD non-blushers, F(1,57) = 9.9, p = .003, j2 = 15, and the difference between the normal control group and the SAD blushers was again borderline significant, F(1,51) = 3.6, p = .063, j2 = 07. However, in contrast to the analysis without sex as covariate the difference between the normal control group and the SAD non-blushers became borderline significant as well, F(1,49) = 3.0, p = .088, j2 = 06. 4 The skin conductance data were, as cheek temperature, influenced by gender. Therefore, the ANOVAs were run again with gender included as a covariate. The results for the ANOVAs with the combined SAD group versus the normal controls remained the same. That is, both a quadratic, F(1,78) = 7.7, p = .007, j2 = .09, as a linear interaction, F(2,77) = 4.6, p = .035, j2 = .06, appeared. This linear interaction was again explained by a greater decrease for the SAD group than for the normal controls. The quadratic interaction was again explained by the normal controls to have a greater increase of skin conductance level from baseline to social task than the SAD group, F(2,77) = 11.4, p = .001, j2 = .13. Moreover, again no significant difference was found for social task to recovery.y including sex as covariate some differences appeared with the original analyses. Not only was the linear interaction explained by a greater decrease in skin conductance level for the SAD blushers compared to the normal control subjects, F(1,50) = 8.9, p = .004, j2 = .15, but also compared to the SAD nonblushers, F(1,56) = 5.2, p = .025, j2 = .09. Moreover, the quadratic interaction was now not only explained by the difference between the normal controls with the SAD nonblushers, F(1,47) = 8.6, p = .005, j2 = .15, but also with the SAD blushers, F(1,50) = 4.9, p = .031, j2 = .09. This indicated that the normal controls were more responsive on skin conductance from baseline to task compared to both SAD groups (respectively SAD blushers: F(1,53) = 10.2, p = .002, j2 = .17; SAD non-blushers: F(1,50) = 8.6, p = .005, j2 = .15). However, the SAD non-blushers recovered less well from social task to recovery compared to both the normal controls, F(1,50) = 8.1, p = .007, j2 = .15, as well as the SAD blushers, F(1,59) = 4.5, p = .038, j2 = .08.

91

quadratic, F(1,79) = 8.2, p = .005, j2 = .09, and a linear interaction, F(1,79) = 5.2, p = .026, j2 = .06. The linear interaction resulted from a general greater decrease for the SAD group than for the normal controls. The quadratic interaction was explained by the normal controls to have a greater increase of skin conductance level from baseline to social task than the SAD group, t(79) = 3.5, p = .001, d = 0.8. No significant difference appeared for social task to recovery. The analyses in which the three groups were included as between subjects variables specified that the linear interaction was explained mainly by a greater decrease in skin conductance level for the SAD blushers compared to the normal control subjects, F(1,51) = 7.4, p = .009, j2 = .13. Moreover, the quadratic interaction was mainly explained by a greater responsiveness to the social tasks of the normal controls compared to the SAD non-blushers, F(1,48) = 11.3, p = .002, j2 = .19. That is, the increase from baseline to social task, F(1,50) = 10.4, p = .002, j2 = .18, and the decrease from social task to recovery was greater for the normal controls than for the SAD non-blushers, F(1,48) = 7.8, p = .008, j2 = .14. The general picture that seemed to appear is that the normal controls exhibited a greater responsiveness to the social tasks than the SAD patients. It is worth noting that the two SAD groups showed a somewhat higher, but non-significant, baseline level of skin conductance than the normal control group (combined SAD group: 2.6, S.D. = .7; normal controls: 2.4, S.D. = .7). This might explain the greater responsiveness of the normal control group compared to the SAD group. 3. Discussion Studies to date are inconclusive with regard to whether SAD patients blush more than others. One reason for these conflicting results might be that SAD patients are a heterogeneous group: those who complain about blushing blush more and/or have a heightened general arousability than those who do not complain about blushing. Indeed, our results showed that SAD patients who identified themselves as blushers showed a different physiological blushing response during social interaction than patients who did not complain about blushing. That is, SAD blushers showed an overall higher cheek temperature and a higher blood flow blushing response and recovery than SAD non-blushers. Compared to controls, SAD blushers had borderline higher overall cheek temperature. However, with regard to blood flow, SAD blushers did not differ in their response or recovery from normal controls. In contrast, SAD non-blushers showed a less intense response and recovery in blood flow to the social challenge than normal controls as well as SAD blushers. In other words, SAD non-blushers showed a ‘suppressed’ blush pattern compared to the other two groups. As for skin conductance level, our results indicate that, in general, SAD patients are not characterized by a heightened social stress response. That they showed a pattern of lower skin conductance response in the current study may have resulted from the somewhat (though not significantly) higher initial (overall) level of skin conductance in both SAD groups (Fig. 1). This may have given the SAD groups less room to increase. It is informative to consider the current findings for the three types of physiological indices (i.e., temperature, blood flow, and skin conductance) in light of prior findings. With regard to cheek temperature, none of three studies conducted thus far was able to differentiate their experimental group from their control group (Edelmann and Baker, 2002; Mulkens et al., 1997, 1999), whereas in the current study SAD blushers showed a higher cheek temperature than SAD non-blushers and a borderline higher temperature than normal controls. Reasons why this was not found in the previously mentioned studies could be that Edelmann and Baker (2002) used a smaller sample (18 SAD patients versus 66 in the current study) and failed to divide their SAD group into

92

M.J. Voncken, S.M. Bo¨gels / Biological Psychology 81 (2009) 86–94

blushers and non-blushers, as the current study did. Moreover, both Mulkens et al. (1997,1999) used an analogue student population instead of a patient sample, as in the current study. Cheek temperature may be higher in a patient population. Another reason why Mulkens et al. (1997,1999) might not have found differences for cheek temperature is that they used another experimental set-up. These studies were specifically aimed at inducing embarrassment, whereas the current study was aimed at inducing social fear or social stress. Hofmann et al. (2006) put forward that social stress brings about a different physiological pattern than embarrassment. Although their physiological data do not support this view, they did not assess cheek temperature but cheek blood flow. It could well be that cheek temperature is an indication of social stress and not of embarrassment. That is, an increase in cheek temperature may be the result of general reddening of the face that occurs by prolonged exposure to social attention (also called a creeping blush) whereas embarrassment is accompanied by short blushes that disappear quickly (Leary et al., 1992). In other words, non-findings in earlier studies about cheek temperature may be explained by not differentiating SAD blushers from SAD non-blushers, not including a patient sample, and inducing embarrassment rather than social stress. Whereas cheek temperature differentiated SAD blushers from SAD non-blushers and normal controls, blood flow showed a different pattern. First, the combined SAD group did not differ in blood flow response to our social tasks compared to the normal control group. This is in line with the results of five previous studies (Drummond, 2001; Drummond et al., 2007; Gerlach et al., 2001; Mulkens et al., 1997, 1999). It is worth noting that although Gerlach et al. (2001) did find differences between their SAD group and the normal control participants during their embarrassment inducing video task they did not find, in line with our finding, differences during the conversation and speech task. In contrast to these five studies, two other studies, using analogue populations, did find their experimental group to show higher blood flow than their normal control group (Bo¨gels et al., 2002; Hofmann et al., 2006). Hofmann et al. (2006) found a greater blood flow response for shy people than for non-shy individuals during an impromptu speech and an embarrassment-inducing singing task. As suggested earlier, blood flow response might indicate embarrassment. Although not intended, the embarrassment of the singing task in Hofmann’s study may have transferred to the speech task, as the subjects participated in the singing task before they conducted the speech. This might also explain why Gerlach et al. (2001) did find their SAD patients to have higher blood flow responses to their embarrassment inducing video task compared to their normal control group but failed to find differences on blood flow for their social stress inducing social performance tasks. The other study, Bo¨gels et al. (2002), did not find differences in blood flow responses but found their socially anxious group to have a higher blood flow level than their non-socially anxious group. It is worth noting that, although it was not significant, SAD patients in this study seemed to have an overall higher blood flow level than normal controls (see Fig. 1). As was argued before, blood flow level is recorded only with large measure error due to differences in skin thickness, placement of the probe on the cheek, etc. Therefore, larger numbers of subjects/larger samples are needed to shed more light on average blood flow levels in SAD patients versus normal controls. Second, several researchers have suggested that SAD patients who complain about blushing mainly have a distorted view concerning their blush but do not show actual physiological blushing problems (see for overview Hofmann et al., 2004). However, the SAD blushers in our study showed a greater blood flow response than the SAD non-blushers; moreover, our cheek temperature data indicated that SAD blushers experience higher cheek temperature than both normal controls and SAD non-

blushers. Apparently, SAD patients who complain about blushing suffer from more physical blushing sensations than SAD patients without blushing complaints. It is also worth noting that the confederates in this study observed higher blushing responses in the SAD blushers compared to the SAD non-blushers and the normal controls. Therefore, although SAD blushers may overestimate their blushing response, they do experience objective physical changes, that others can observe. The third finding regarding blood flow was that, SAD nonblushers showed a ‘suppressed’ blush pattern compared to the other two groups. To our knowledge, the only study that directly compared SAD blushers with SAD non-blushers in blood flow responses is that of Gerlach et al. (2001). They did not find differences in blood flow responses between these two groups. Power problems in their study (n = 15 per group) might explain their non-finding. The current study increased its power by averaging two social tasks and the two blood flow indices and by using a larger patient population (32 SAD blushers and 34 SAD non-blushers). This less intense blush pattern among SAD nonblushers was not only evident for blood flow but also for cheek temperature. That is, SAD non-blushers showed a borderline lower cheek temperature than the normal controls and a significantly lower cheek temperature than SAD blushers. We are not aware of any other studies with comparable findings. Three explanations might explain these findings. First, on a psychophysiological level this ‘suppressed’ pattern might be explained as follows. The sympathetic reaction to stress or anxiety is a vasoconstriction of blood vessels (Leary et al., 1992), thus a decrease in blood flow in the face. Although it is speculative, it might be that both the SAD blushers as the SAD non-blushers show this vasoconstriction of blood vessels as they experienced social stress during the social tasks. This might explain why the SAD non-blushers showed a lower blood flow response than the normal control participants. However, in the SAD blushers, this longer lasting vasoconstriction of blood vessels might be counteracted by numerous short-lived blush responses which are accompanied by vasodilatation of blood vessels (i.e., an increase of blood flow). This might be the reason why the SAD blushers showed a blood flow pattern similar to the normal control participants. Second, the suppressed physiological blushing pattern in SAD non-blushers may also result from being more inhibited (i.e., the tendency to avoid unfamiliar events and places) than SAD blushers. Inhibited individuals have been shown to have a lower physiological reactivity than uninhibited individuals (see for a theoretical framework Kagan et al., 1988 and for supporting evidence Woodward et al., 2000 or Schwartz et al., 1999). Hofmann et al. (2004) also proposed a ‘shyness’ or ‘inhibited’ subtype in SAD. For purpose of another study (Voncken et al., 2008), we had data on how many words were said during the getting acquainted conversation. This data revealed SAD non-blushers were less able to elicit conversation from confederates than SAD blushers. That is, confederates spoke fewer words during the conversation with SAD non-blushers than with SAD blushers (t(57) = 2.0, p = .049). This suggests that SAD non-blushers may, indeed, represent inhibited individuals, which is linked with an inhibited psychological response pattern. Clearly, as the current study did not aim to investigate an inhibited subtype in SAD, further studies are needed to investigate this possibility. Third, the most speculative explanation for a less responsive pattern for SAD non-blushers might be that these people suppress their blushing response by worrying. Within the worry literature it is suggested that worrying suppresses physiological responses (see for a review Borkovec and Roemer, 1995; Hofmann et al., 2004. Interestingly, life-time depressive disorders were significantly more prevalent among SAD non-blushers than among SAD

M.J. Voncken, S.M. Bo¨gels / Biological Psychology 81 (2009) 86–94

blushers. Watkins and Moulds (2007) concluded that the same style of thinking, as described in worriers, is present in patients with major depression. Thus, although speculative, a worry-type style of thinking in the SAD non-blushers might explain the suppressed blush reaction pattern in these patients. It is generally believed that people cannot estimate their own blushing response well (see for an overview Drummond et al., 2007). Interestingly, this study was the first to find correlations in healthy participants between cheek blood flow and (borderline significant) cheek temperature on one hand and self-report rating of blushing on the other hand. This would indicate that people without SAD use physiological changes in their cheeks to estimate the frequency and intensity of their blushing. To a lesser extent SAD blushers also seem able to detect their blush. That is, a borderline significant correlation was found between SAD blushers’ self-report blush and cheek blood flow. However, for SAD non-blushers the normal detecting of own blushing seems disrupted as no correlations were found between the physiological measures of blushing and self-report. One tentative explanation is that they are not aware of their ‘‘suppressed’’ blushing response. Some remarks need to be made with regard to the different physiological assessments of blushing. In line with the suggestions of Drummond (2001), the physiological responses measured with cheek and forehead blood flow appeared to reflect the same underlying physiological changes. That is, not only did these two measures correlate well, but also the patterns across the three groups regarding these indices were highly similar. It is worth noting that the correlations for the recovery phase between cheek temperature and blood flow were low. This might be explained as follows. A temperature probe shows latency in response (see Schandry and Poth, 1983, who estimate a latency of 10–30 s), as the probe itself needs to cool down before a decrease in temperature is recorded. A photoplethysmograph, on the other hand, is a recording device that picks up the reflection of an infrared light and can thus pick up changes more easily. Therefore, the low correlations between blood flow and temperature in the recovery phase probably reflected differences in the recording devices. As for skin conductance level, several studies failed to detect a higher skin conductance level in socially anxious individuals during social tasks (Edelmann and Baker, 2002; Gerlach et al., 2001; Bo¨gels et al., 2002; Mulkens et al., 1997, 1999). Our results are in line with these studies. In addition, in contrast to the idea of Hofmann et al. (2004) that SAD blushers correspond to the fearful subtype (i.e., with a heightened stress response) of SAD, we did not find SAD blushers to be characterized by a general heightened stress response. Note, however, that we measures skin conductance and not heart rate, another physiological indicator of general fearfulness. Skin conductance level did not relate in any way to the three physiological indices of blushing. Shearn et al. (1990) also found no relation between skin conductance and blushing measures during events intended to evoke a blush. Moreover, although social tasks, in general, evoke stress (Beidel et al., 1985; Hofmann et al., 1995), skin conductance did not differentiate SAD patients from others in prior studies (Edelmann and Baker, 2002; Gerlach et al., 2001; Mulkens et al., 1997, 1999). Thus, it seems that the blushing response during social stress arises from a different physiological system than general arousal when measured by skin conductance. The current study had some limitations. Artificial social interactions were used, and it is necessary to see whether such findings translate to real-life situations. Assessments were restricted to a single speech and conversation situation, whereas people with SAD tend to have problems in a greater variety of social interactions. Moreover, our tasks aimed to induce social stress and not embarrassment. Therefore, we were not able to

93

assess differences between the groups on embarrassment-induced blushing. The gender ratio between the two SAD groups was not equal, that is, the SAD blushers consisted of more females than the SAD non-blushers. Although gender did not affect the blood flow responses and the analyses in which we corrected for gender (i.e., temperature and skin conductance analyses, see footnotes 3 and 4) showed in general the same patterns, we cannot be sure that gender did not influence some of our results. The results of this study have some implications for treatment. It seems that SAD patients who report blushing problems have more physiological blushing responses than those who do not report these problems, and their blush is also detected by independent raters. This may indicate that SAD blushers focus more attention on their blushing response, not only because of their fear of blushing itself, but also because of an actual heightened blushing response. As self-focused attention on one’s blush may maintain and exacerbate blushing, decreasing selffocused attention would be an important intervention for this group. Bo¨gels (2006) concludes that task concentration training, in combination with cognitive treatment, is highly effective for this subgroup of SAD patients who complain about blushing. Another study by Bo¨gels and Voncken (2008) shows that this subgroup also benefits from a specific type of skills treatment in which they learn to continue to behave normally despite blushing responses and to cope with blushing problems by, for example, making small jokes about their blush. With regard to SAD non-blushers, it may be that they are more inhibited individuals. If so, they might profit from treatment in which time is spent improving interpersonal contact, for example, by being more open through self-disclosure in order to elicit more conversation. Acknowledgments This research was supported by a grant from the Netherlands Organization for Scientific Research (NWO: 015.000.069) to the second author. Our gratefulness goes out to all the patients that were willing to participate in this anxiety provoking study and to all the confederates. We would like to thank the experimenters Serife Alakir, Eline Smit, Anja Hendriks and Saskia Nijst and the research assistants Thamare van Roosmalen, Philippe Jacques and Esin Demir. In addition, we acknowledge Erik Schouten for statistical help and Theo Aerts for his help with the equipment of the laboratory. Moreover, we would like to thank all the therapists of the anxiety program at the RIAGG Maastricht, especially Guido Sijbers, Stefanie Duijvis, Rene Albers and Hannie van Genderen. Arnoud Arntz is thanked for his useful comments on this manuscript. References Beidel, D.C., Turner, S.M., Dancu, C.V., 1985. Physiological, cognitive and behavioral aspects of social anxiety. Behaviour Research and Therapy 23, 109–117. Bo¨gels, S.M., Stein, M., 2009. Social phobia: Moving to DSM V. In: G. Andrews, D.S. Charney, P.J. Sirovatka, D. A. Regier (Eds.) Stress-induced and fear circuitry disorders: Advancing the research agenda for DSM V, Arlington, VA: American Psychiatric Association, pp. 65–83. Bo¨gels, S.M., 2006. Task concentration training versus applied relaxation, in combination with cognitive therapy, for social phobia patients with fear of blushing, trembling, and sweating. Behaviour Research and Therapy 44, 1199–1210. Bo¨gels, S.M., Mansell, W., 2004. Attention processes in the maintenance and treatment of social phobia: hypervigilance, avoidance and self-focused attention. Clinical Psychology Review 24, 827–856. Bo¨gels, S.M., Mulkens, S., Jong de, P.J., 1997. Task concentration training and fear of blushing. Clinical Psychology and Psychotherapy 4, 251–258. Bo¨gels, S.M., Reith, W., 1999. Validity of two questionnaires to assess social fears: The Dutch Social Phobia and Anxiety Inventory and the Blushing, Trembling and Sweating Questionnaire. Journal of Psychopathology and Behavioral Assessment 21, 51–66. Bo¨gels, S.M., Rijsemus, W., De Jong, P.J., 2002. Self-focused attention and social anxiety: the effects of experimentally heightened self-awareness on fear, blushing, cognitions, and social skills. Cognitive Therapy and Research 26, 461–472.

94

M.J. Voncken, S.M. Bo¨gels / Biological Psychology 81 (2009) 86–94

Bo¨gels, S.M., Voncken, M.J., 2008. Social skills training versus cognitive therapy for social anxiety disorder characterized by fear of blushing trembling or sweating. International Journal of Cognitive Therapy 1, 138–150. Boone, M.L., McNeil, D.W., Masia, C.L., Turk, C.L., Carter, L.E., Ries, B.J., Lewin, M.R., 1999. Multimodal comparisons of social phobia subtypes and avoidant personality disorder. Journal of Anxiety Disorders 13, 271–292. Borkovec, T.D., Roemer, L., 1995. Perceived functions of worry among generalized anxiety disorder subjects: distraction from more emotionally distressing topics? Journal of Behavior Therapy and Experimental Psychiatry 26, 25–30. Darwin, C., 1955. The Expression of the Emotions in Man and Animals. Greenwood Press, New York. Drummond, P.D., 2001. The effect of true and false feedback on blushing in women. Personality and Individual Differences 30, 1329–1343. Drummond, P.D., Back, K., Harrison, J., Helgadottir, F.D., Lange, B., Lee, C., Leavy, K., Novatscou, C., Orner, A., Pham, H., Prance, J., Radford, D., Wheatley, L., 2007. Blushing during social interactions in people with a fear of blushing. Behaviour Research and Therapy 45, 1601–1608. Drummond, P.D., Lim, H.K., 2000. The significance of blushing for fair- and darkskinned people. Personality and Individual Differences 29, 1123–1132. Edelmann, R.J., 1990. Chronic blushing, self-consciousness, and social anxiety. Journal of Psychopathology and Behavioral Assessment 12, 119–127. Edelmann, R.J., Baker, S.R., 2002. Self-reported and actual physiological responses in social phobia 41, 1–14. First, M.B., Spitzer, R.L., Gibbon, M., Williams, G.B.W., Benjamin, L., 1997. Structured Clinical Interview for DSM-IV Personality Disorders (SCID-II). American Psychiatric Press, Washington, DC. Gerlach, A.L., Wilhelm, F.H., Gruber, K., Roth, W.T., 2001. Blushing and physiological arousability in social phobia. Journal of Abnormal Psychology 110, 247–258. Hofmann, S.G., Heinrichs, N., Moscovitch, D., 2004. The nature and expression of social phobia: toward a new classification. Clinical Psychology Review 24, 769– 797. Hofmann, S.G., Moscovitch, D.A., Kim, H.J., 2006. Autonomic correlates of social anxiety and embarrassment in shy and non-shy individuals. International Journal of Psychophysiology 61, 134–142. Hofmann, S.G., Newman, M.G., Ehlers, A., Roth, W.T., 1995. Psychophysiological differences between subgroups of social phobia. Journal of Abnormal Psychology 104, 224–231. Kagan, J., Reznick, J.S., Snidman, N., 1988. The physiology and psychology of behavioral inhibition in children. Annual Progress in Child Psychiatry and Child Development. Laederach Hofman, K., Mussgay, L., Buchel, B., Widler, P., Ruddel, H., 2002. Patients with erythrophobia (fear of blushing) show abnormal autonomic regulation in mental stress conditions. Psychosomatic Medicine 64, 358–365.

Leary, M.R., Britt, T.W., Cutlip, W.D., Templeton, J.L., 1992. Social blushing. Psychological Bulletin 112, 446–460. Leary, M.R., Meadows, S., 1991. Predictors, elicitors, and concomitants of social blushing. Journal of Personality and Social Psychology 60, 254–262. Mersch, P.P., Hildebrand, M., Lavy, E.H., Wessel, I., et al., 1992. Somatic symptoms in social phobia: a treatment method based on rational emotive therapy and paradoxical interventions. Journal of Behavior Therapy and Experimental Psychiatry 23, 199–211. Mulkens, S., De Jong, P.J., Bo¨gels, S.M., 1997. High blushing propensity: fearful preoccupation or facial coloration? Personality and Individual Differences 22, 817–824. Mulkens, S., de Jong, P.J., Dobbelaar, A., Bo¨gels, S.M., 1999. Fear of blushing: fearful preoccupation irrespective of facial coloration. Behaviour Research and Therapy 37, 1119–1128. ¨ Ost, L.G., Jerremalm, A., Johansson, J., 1981. Individual response patterns and the effects of different behavioral methods in the treatment of social phobia. Behaviour Research and Therapy 19, 1–16. Scholing, A., Emmelkamp, P.M., 1993. Cognitive and behavioural treatments of fear of blushing, sweating or trembling. Behaviour Research and Therapy 31, 155– 170. Schandry, R., Poth, E., 1983. Eine experimentelle Untersuchung zur Psychologie und Physiologie des Errotens/An experimental investigation on the psychology and physiology of blushing. Psychologische Beitrage 25, 503–514. Schwartz, C.E., Snidman, N., Kagan, J., 1999. Adolescent social anxiety as an outcome of inhibited temperament in childhood. Journal of the American Academy of Child and Adolescent Psychiatry 38, 1008–1015. Shearn, D., Bergman, E., Hill, K., Abel, A., et al., 1990. Facial coloration and temperature responses in blushing. Psychophysiology 27, 687–693. Turner, S.M., Beidel, D.C., Dancu, C.V., Stanley, M.A., 1989. An empirically derived inventory to measure social fears and anxiety: The Social Phobia and Anxiety Inventory. Psychological Assessment 1, 35–40. Voncken, M.J., Alden, L.E., Bo¨gels, S.M., Roelofs, J., 2008. Social rejection in social anxiety disorder: the role of performance deficits evoked negative emotions and dissimilarity. British Journal of Clinical Psychology 47, 439–450. Voncken, M.J., Bo¨gels, M.S., 2008. Social performance deficits in social anxiety disorder: reality during conversation and biased perception during speech. Journal of Anxiety Disorders 22, 1384–1392. Watkins, E., Moulds, M.L., 2007. Reduced concreteness of rumination in depression: a pilot study. Personality and Individual Differences 43, 1386–1395. Woodward, S.A., Lenzenweger, M.F., Kagan, J., Snidman, N., Arcus, D., 2000. Taxonic structure of infant reactivity: evidence from a taxometric perspective. Psychological Science 11, 296–301.